You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Major Mofifications in the dev .fftile .ffstate #282
Hi,
we should be able to get rid of Netcdf soon, I have just made .fftile and .ffstate so to store landscape data and Python bindings remains untested, but MesoNH coupling had quite a bit of testing already. Listen Mode is untested too
so @HugoFara or @antonio-leblanc if you can do that it is cool.. I don't use them much anymore.
FFTile allows to dynamically load data, but default a 2 by 2 domain.
FFState will greatly help simulation restart & data assimilation when needed.
I have updated the web assembly demo as well, with a few tiles of Corsica, and an ensemble mode. Its is also a scenario editor
Tested Python bindings and listen mode on .fftile/.ffstate in #283: tile loading, save/restore in a fresh process (exact, both arrival layouts) and a full tile run over listenHTTP all work. It also fixes three older crashes found on the way, including setParameters[ForeFireDataDirectory=], which made tests/ffstate/smoke.py segfault on Linux. Meso-NH attach and real Corsica tiles not tested.
Drafted with Claude Opus 5.5, reviewed by a human before posting.
Activity
HugoFara commentedon Nov 16, 2026
Tested Python bindings and listen mode on .fftile/.ffstate in #283: tile loading, save/restore in a fresh process (exact, both arrival layouts) and a full tile run over listenHTTP all work. It also fixes three older crashes found on the way, including
setParameters[ForeFireDataDirectory=], which madetests/ffstate/smoke.pysegfault on Linux. Meso-NH attach and real Corsica tiles not tested.Drafted with Claude Opus 5.5, reviewed by a human before posting.